Energy Harvesting in 1D Photonic Multilayer Solar Cells

integrat-ing sphere or inside the integratintegrat-ing sphere (cf. section 3.1). Controlled manipulations of the overall device absorption of thin-lm solar cells featuring either a 1D or 2D photonic structure are investigated in the following sections.

4.1. Energy Harvesting in 1D Photonic Multilayer Solar Cells

The following section is based on the research paper Decoupling Optical and Elec-tronic Optimization of Organic Solar Cells using High-Performance Temperature-Stable TiO₂/Ag/TiO₂ Electrodes which was written as part of my Ph.D.[153] All experiments and measurements for this study were performed by Dr. Kwang-Dea Kim. I designed the project, helped with simulations and interpretation of the data and wrote most parts of the manuscript.

4.1.1. Device preparation

ITO-free electrode fabrication. The TiO2/Ag/TiO2 (TAT) multilayer was sequen-tially deposited onto glass substrate (PGO) by dc sputtering of a TiO₂ target (99.99 % purity) and Ag target (99.99 % purity) at room temperature in a high vacuum system with a base pressure of 3·10^-7 Torr. TiO2 and Ag were deposited using dc power of 100 W. The thicknesses of the TiO₂ layer and Ag were calibrated using spectroscopic ellipsometry and atomic force microscopy (section 3.1), respectively.

OPV device fabrication: To fabricate the inverted OPV devices, layers of a donor-acceptor blend, PEDOT:PSS, and Ag top-electrode were sequentially deposited onto TiO₂-coated ITO glass and TAT-coated glass (14×14 mm²) which were annealed at 400^◦C for 30 min under vacuum condition (5·10^-3 Torr). For the preparation of the active layer, PTB7 (1-Material) and PC71BM (Solenne) (8:12 mg) were dissolved in chlorobenzene/1,8-diiodooxtane (CB/DIO; 0.97/0.03 ml, Aldrich). Both donor and ac-ceptor materials are used without further purication. The solution was stirred at45^◦C

4.1 Energy Harvesting in 1D Photonic Multilayer Solar Cells

overnight. The active layer was deposited onto the TiO₂-coated ITO glass (Lumtec) and TiO₂/Ag/TiO₂ coated glass substrate using spin-coating at 1000 rpm for 40 s un-der N2 atmosphere. For the preparation of the PEDOT:PSS (Clevios P VP AI 4083, Heraeus) solution, PEDOT:PSS and IPA were mixed in a ratio of 1:10. This solution was spin-coated onto the active layer at 5000 rpm for 60 s under N₂ condition. The 150 nm Ag top electrode was thermally evaporated onto the PEDOT:PSS layer at an evaporation rate of 1 Å s^-1 at a pressure of 5·10^-6Torr.

4.1.2. Decoupling Optical and Electronic Optimization of Organic Solar Cells using High-Performance Temperature-Stable TiO2/Ag/TiO2

Electrodes

Organic photovoltaic (OPV) devices have attracted much interest during the last two decades, due to the possibility of low-cost fabrication, lightweight, exibility, and sim-ple fabrication processing.[23, 154156] Power conversion eciencies up to 10 % have been reported recently,[157, 158] even though this is still inferior to the theoretically predicted 20 % eciency for organic single junction devices.[159] OPVs are classied within the distinct class of excitonic solar cells, i.e., coulombically bound electron-hole pairs are generated upon light-absorption with binding energies exceeding the thermal energy (k_BT) at room temperature. The electrochemical driving-force given at a type II staggered heterointerface is necessary for free charge-carrier generation.[20] Therefore, only excitons generated within the exciton diusion length (typically around 10 nm) towards a donor-acceptor interface are successfully harvested. Internal quantum e-ciencies (IQEs) up to 100 % have been reported.[35] In case of the IQE being unity every absorbed photon, i.e., generated exciton is separated and all photogenerated po-larons are extracted from the active layer and collected at the external electrodes of the device.[18, 35, 160162] Electronically optimized organic thin lms with suitable donor-acceptor macro phase separation exhibiting such high IQE can only be realized using lm thicknesses around 100 nm, which results in severe performance losses due to limited light absorption.[67, 160] In contrast, thicker layers exhibiting virtually com-plete photon harvesting, but suer from reduced IQEs owing to pronounced charge carrier recombination losses. This trade-o motivates researcher to introduce light management structures into the photocurrent generating layers that either localize the electro-magnetic energy in the near-eld of plasmonic nanostructures or increase the optical path length due to scattering into lateral modes.[163, 164] One requirement for their successful implementation is that the structural changes due to the light manage-ment structures do not inuence the electronic properties of the organic layer, which is crucial especially for spontaneously intermixed BHJ devices. Therefore, it is necessary to introduce a simple and suitable device design ideally featuring at interfaces in or-der to allow for decoupling of optical and electronic optimization. Here, we focus on the tuneability of the coherent electro-magnetic eld distribution in at interface OPVs that naturally dene an optical cavity (Figure 4.1a). The tuneability of the cavity is accessible by replacing the commonly used ITO with a TiO₂/Ag/TiO₂(TAT) sandwich structure and a variation of the respective layer thicknesses. In particular, variation

Figure 4.1.: Schematic device structure and characterization of device performance and op-tical properties. (a) Schematic view of the fabricated photovoltaic cells, picture of TiO₂/Ag/TiO₂ (TAT): 20/12/28 nm (ITO-free) electrode, and simulation of optical electric eld proles in terms of the normalized intensity (|E|²) depending on the bottom TiO₂ thickness (28 nm; red, 40 nm;

blue and 55 nm; yellow); here, the wavelength of 550 nm was chosen. (b) Current-voltage (J-V) characteristics of OPVs with T/ITO and TAT (ITO-free) electrodes. (c) Experimentally measured and (d) simulated EQE spectra of devices with T/ITO and TAT (ITO-free) electrodes. (e) Relative EQE and absorption of (TAT-T/ITO)/TAT. Reprinted with permission from Kim and Pfadler et al.[153] Copyright (2015), AIP Publishing LLC.

of the thickness of the bottom TiO2 layer does not inuence the electronic proper-ties of the electrode since the charge collecting second TiO₂ layer, which is in contact with the organic active layer, remains unchanged. In general, high transparency of the transparent conducting electrode (TCE) is of great importance for the eective light ab-sorption in the device, which is nalized by an Ag/Al backelectrode (backmirror).[165 167] ITO has been commonly used for various optoelectronic devices as transparent electrode due to its excellent transparency and electrical properties. However, the price of ITO is rising due to the limited availability of indium.[168, 169] Besides, ITO deposition requires high-temperature vacuum processing and ITO has a low thermal stability caused by ion diusion at temperatures exceeding 300^◦C.[170] The thickness of the ITO layer determines its properties (electronic and optical transmission). ITO thin lms have a xed thickness of around 200 nm for optimum performance for solar

4.1 Energy Harvesting in 1D Photonic Multilayer Solar Cells

cell applications. These drawbacks of ITO are the driving force for researchers to in-vestigate alternative materials - a number of promising candidates have been already identied. One class are doped metal oxides such as Al- and Ga-doped ZnO (AZO and GZO), which are cheaper than ITO.[167] Additionally, the use of conducting car-bon materials including carcar-bon nanotubes, graphene, and conducting polymers such as poly(3,4-ethylenedioxythiophene):poly-(styrenesulfonate) (PEDOT:PSS) has been pro-posed, which allows solution-processing of the transparent electrode.[171, 172] How-ever, most alternative materials are inadequate as replacement for ITO due to their lower optical transparency and/or conductivity in comparison to ITO. Recently, ox-ide/metal/oxide (OMO) multilayer structures featuring very thin metal layers (Ag, Au, or Cu) sandwiched between two metal oxides (TiO₂, ZnO, or MoO₃) have been demonstrated as TCEs.[166, 173175] Sandwich structures based on TiO2 are promis-ing candidates due to their transparency in the visible, and the strong mechanical and chemical stability of TiO₂.[176] Dhar et al. demonstrated TAT multilayer electrodes with good optoelectronic properties exhibiting a sheet resistance (Rsheet) of 5.7 Ω ^-1 and an average optical transmittance of 90 % at 590 nm.[173] However, these structures have not been successfully applied to thin-lm devices like OPVs, dye-sensitized solar cells (DSSC), or the recently emerging perovskite solar cells.

In this section we demonstrate high-performance state of the art OPVs using TAT mul-tilayers as an ITO-free electrode which allows for single junction eciencies up to 8.7 %.

The possibility to individually control the lm thicknesses in the multilayer allows for an optimization of the light intensity prole in the active layer as a function of the pho-toactive material, which directly translates into higher photocurrents and more ecient devices.

Results and Discussion. ITO-free transparent electrodes with a TAT multilayer struc-ture on glass were prepared by sputtering at room temperastruc-ture without any vacuum break (see Figure 4.1a). To fabricate these OPVs, the ITO-free electrode consisting of the top TiO₂ layer as an electron collector was annealed at 400^◦C for 30 min. This thermal processing is crucial for charge collection, directly evident from current density-voltage curves, mostly reected in improved ll factors. Interestingly, the R_sheet of TAT electrodes is stable after annealing even at elevated temperatures up to 550^◦C with slight increase in Rsheet value of only 1.69 Ω ^-1 (from 6.75 to 8.44 Ω ^-1), whereas the R_sheet of TiO₂-coated ITO (T/ITO) annealed at the same condition was signif-icantly increased from 11.23 to 21.65 Ω ^-1 (Figure 4.2). This indicates that the TAT electrodes have a high thermal stability and maintain their good electrical prop-erties. In contrast, upon the same thermal treatments ITO electrodes show signi-cantly deteriorating performance. Owing to this low thermal stability, ITO has been narrowly used only for the application of electric devices fabricated at temperatures below 300^◦C.[170] The advantage of better thermal stability of our TAT electrodes provides a wide opportunity in this eld and makes them viable for a number of de-vices based on TiO2 electrodes like DSSCs, perovskite photovoltaics, and other hybrid inorganic-organic solar cells. Besides the superior thermal stability of TAT electrodes in comparison to ITO electrodes they have the advantage that the absorption behaviour,

Figure 4.2.: Thermal stability test of T/ITO and TAT electrodes. Sheet resistance (R_sh) of T/ITO and TAT (20/12/28 nm) electrodes depending on dierent annealing temperature at 350, 400, 450, 500 and 550^◦C for 30 min under vacuum condition of 3·10^-3 Torr. Reprinted with permission from Kim and Pfadler et al.[153] Copyright (2015), AIP Publishing LLC.

i.e., the coherent electric eld intensity (|E|²) distribution inside the active layer of the device can be maximized by varying the thickness of the TiO2 layers without the danger of structure-induced changes to the active layer morphology. Therefore, match-ing the optical cavity for an optimized active material processmatch-ing parameter like the active layer thickness results in a maximized device performance due to enhanced light harvesting, while keeping the electronic properties of the organic layer completely un-aected. The computational results in Figure 4.1a show the coherent electric eld proles inside OPV devices as a function of the bottom-TiO2 layer thickness. Opti-cal simulations based on a transfer matrix algorithm (section 3.2) show that changes in thickness of the individual layers crucially inuences the appearance of the optical cavity mode.[124] Figure 4.1b shows the current density-voltage (J-V) characteristics of poly[[4,8-bis[(2-ethylhexyl)oxy]benzo[1,2-b:4,5-b']dithiophene-2,6-diyl][3-uoro-2-[(2-ethylhexyl)carbonyl] thieno[3,4-b]thiophenediyl]] (PTB7): [6,6]-phenyl C71-butyric acid methyl ester (PC71BM) OPVs fabricated with two dierent electrodes, namely T/ITO and TAT multilayer. Using the TAT: 20/12/28 nm multilayer, a representative power conversion eciency (PCE) of 8.7 % (ll factor, FF of 65.8 % and open circuit volt-age, VOC of 0.76 V) is reached with a signicantly higher short-circuit current density (J_SC) of 17.54 mA cm^-2. This is a relative enhancement of 12 % in J_SC ([J_SC ,TAT-J_SC,T/ITO]/J_SC,T/ITO) with slightly increased FF and V_OC in comparison to a repre-sentative ITO-based device with a JSCof 15.64 mA cm^-2(7.5 % overall eciency, FF of 64.3 % and V_OCof 0.75 V). In order to determine the origin of the superior performance, experimental external quantum eciencies (EQEs) for two dierent OPVs featuring T/ITO and TAT multilayer electrodes are compared. As shown in Figure 4.1c the EQE in the wavelength range of 450-800 nm (i.e., the main absorption of the PTB7:PC₇₁BM blend) with the TAT electrode is higher than with ITO, while there is a slight reduction

4.1 Energy Harvesting in 1D Photonic Multilayer Solar Cells

in the EQE observed between 350 and 450 nm. These ndings are in good agreement with simulated EQE spectra as evident from Figure 4.1d. To verify the decoupling of the optical absorption and the IQE in the measured EQE data, relative EQE and relative total absorption calculated as (TAT-T/ITO)/TAT are compared (Figure 4.1e).

Good agreement between relative EQE and relative total absorption was obtained, indi-cating that the IQE of the active lm remains unchanged and the improvement in EQE can be directly attributed to the increase in light absorption. Figure 4.3 shows the

Figure 4.3.: Electric eld prole inside the OPV system. Simulated result of the optical electric eld prole for OPVs structured as Ag (150 nm)/PEDOT (8 nm)/PTB7-PC71BM (95 nm) on TiO2

(20 nm)/ITO (170 nm) and TiO2 (20 nm)/Ag (12 nm)/TiO2 (28 nm) shown in Figure 4.1b de-pending on the wavelength of incident light (450, 550, 650 and 750 nm). Reprinted with permission from Kim and Pfadler et al.[153] Copyright (2015), AIP Publishing LLC.

simulated optical electric eld proles in the TAT and T/ITO devices from Figure 4.1b as a function of the wavelength of incident light ranging from 450 nm to 750 nm. The full optical electric eld proles ranging from 450-850 nm are presented in Figure 4.4.

The simulations show that a higher intensity (i.e.,|E|²) for wavelengths above 450 nm can be obtained in OPVs with TAT electrode in comparison to the T/ITO reference.

Accordingly, light absorption is enhanced in this wavelength region, resulting in an improvement of device performance due to enhanced photocurrent generation. While

Figure 4.4.: Optical electric eld prole inside the OPV system. Simulated result of the optical electrical eld prole for OPVs structured as Ag (150 nm)/PEDOT (8 nm)/PTB7-PC₇₁BM (95 nm) on (a) TiO2 (20 nm)/ITO (170 nm) and (b) TiO2 (20 nm)/Ag (12 nm)/TiO2 (28 nm) shown in Figure 4.1b depending on the wavelength of incident light (450-800 nm). Reprinted with permission from Kim and Pfadler et al.[153] Copyright (2015), AIP Publishing LLC.

these are the representative results for one dened set of layer thicknesses of the TAT electrode, the optical electric eld can be greatly inuenced by the thickness of each layer in the sandwich electrode. This allows adjusting the TAT electrode to arbitrary active layers with dierent optimized lm thicknesses and/or intrinsic optical proper-ties. In other words, the presented TAT electrodes allow the tailored optimization of OPVs through the control of thickness of the TiO₂ layers in the TAT multilayer system.

There, the optimum condition for the TAT electrode can be predicted using optical modelling of the respective photoactive lm. In order to outline this methodology Figure 4.5 summarizes J_SC simulations for OPVs structured as Ag (150 nm)/PEDOT (8 nm)/active layer on TiO₂/ITO or TiO₂/Ag/TiO₂ with dependencies of the layer

4.1 Energy Harvesting in 1D Photonic Multilayer Solar Cells

Figure 4.5.: Photocurrent simulation. Contour plot of simulated photocurrent of OPV structured as Ag (150 nm)/PEDOT (8 nm)/active layer on TiO₂/ITO (170 nm) and TiO₂/Ag (12 nm)/TiO₂ with change in the thickness of (a) and (c) ITO (x)-TiO₂ (y) and (b) and (d) bottom (x) top (y) TiO₂, respectively. The simulation of two dierent active layer system which are (a) and (b) PTB7:PC₇₁BM (95 nm) and (c) and (d) P3HT:PC₆₁BM (300 nm) was obtained. The thickness of Ag was xed to 12 nm in this simulation. Reprinted with permission from Kim and Pfadler et al.[153] Copyright (2015), AIP Publishing LLC.

thickness of ITO (x) vs TiO₂ (y) and bottom (x) vs top (y) of TiO₂, respectively. The contour plots of J_SCfor OPVs with active layers of 95 nm of PTB7:PC₇₁BM (Figure 4.5a and b) and 300 nm of P3HT:PC61BM (Figure 4.5a and b), representing two dierent examples of electronically optimized OPV active layers, show a pronounced current density-dependency on the layer thicknesses. This result implies that the optimum con-dition for high-performance can be eectively discovered by controlling the thickness of each layer in the simulation for dierent types of photovoltaic devices with dierent absorption spectra of the photoactive materials and dierent active layer thicknesses.

The highest JSC value in each condition is indicated by a red arrow. In the OPV device with PTB7:PC₇₁BM on ITO (Figure 4.5a), the maximum J_SC value of 18.5 mA cm^-2 is observed at the thickness condition of TITO = 60 nm and TTiO₂ = 70 nm. In

contrast, the OPV with exactly the same active layer but on a TAT electrode shows a highest J_SC value of 17.5 mA cm^-2 (T_bottom-TiO2 = 20 nm, 12 nm Ag and T_top-TiO2

= 90 nm). While these results indicate slightly higher possible photocurrents in the ITO system it should be noted that the fabrication of high-performance devices using such thicknesses is challenging because not only optical but also electrical properties of the multilayer devices should be considered. In the case of ITO, the resistance is exponentially increased as the lm thickness is decreased. For instance, high sheet resistances (> R _sheet = 50 Ω ^-1) have been reported for thicknesses below 100 nm, which would directly result in reduced FFs.[177] Hence, the optimum condition from optical simulations would not allow for achieving a high-performance device in case of the ITO electrode. In contrast, the conductivity of the TAT electrode is mostly pro-vided by the thin Ag lm, and the sheet resistance is only marginally inuenced by the thickness of the top and bottom-TiO₂ lm (Figure 4.4). Furthermore, the simulations

Figure 4.6.: Simulated and experimental JSCwith PCE of device. Simulated JSCprole of TAT multilayer with xed Ag/top-TiO2 layer thickness of 12/20 nm (blue line) and experimental JSC

(open square) with changing in the thickness of bottom-TiO2layer (5, 10, 28, 40 and 55 nm). Inset:

Power conversion eciency (PCE). The dotted line in Figure 4.5b represents this simulation result.

Reprinted with permission from Kim and Pfadler et al.[153] Copyright (2015), AIP Publishing LLC.

underline how important it is to optimize the active layer thickness to gain a maximum photocurrent generation in dierent active layer systems as shown for the two examples of PTB7- and P3HT-based OPVs. While commercial ITO substrates are typically fab-ricated at a xed layer thickness at which balanced sheet resistance and transitivity are obtained, the TAT electrodes allow to individually adjust the layer thicknesses in order to maximize light absorption in the active layer. As a proof of concept, the decoupled optical cavity optimization of devices equipped with the TAT multilayer electrodes is exemplarily performed, both by experiment and simulation. For this purpose only the bottom TiO₂ layer is varied in order to keep the thickness of the top layer comparable

4.2 Light-coupling and Light-Trapping in Nanostructured Thin-Film Solar Cells featuring a 2D Photonic Structure

for the ITO and the TAT lm. As found in optimization experiments the thickness of the top TiO₂ layer inuences the electronic properties of the device, mostly due to an increased series resistance and lm roughness for thicker layers. While this issue requires further optimization of the TiO₂ processing in the future in order to enable even wider tunability, the fundamental principle presented in this study is outlined for the exclusive variation of the bottom TiO2 layer. The resulting experimentally obtained JSC values are compared to the simulated values (Figure 4.6). T_bottom-TiO2 thicknesses of 5 and 10 nm resulted in J_SC values of 17.08±0.22 and 16.89±0.23 mA cm^-2, respectively.

The highest JSC value was 17.78±0.31 mA cm^-2 for Tbottom-TiO2 of 28 nm, in good agreement with the simulated curves. By increasing T_bottom-TiO2 from 40 to 55 nm, the J_SC values were gradually decreased from 16.30±0.28 to 15.49±0.80 mA cm^-2. Detailed information for the device performance is presented in Table A1. We note that the experimentally measured J_SC value has a similar trend as the simulated J_SC (≥T_bottom-TiO2 of 10 nm), although the J_SC value in experimental results was slightly higher than the values obtained from simulations. In addition, the PCE follows the JSC

tendency, as shown in Figure 4.6 (inset), underlining that the electronic properties of the active layer are not aected by tuning the TiO₂ bottom layer. This further implies that the change in thickness of the bottom-TiO2 layer is exclusively responsible for the device performance as it directly determines the coherent electric eld distribution in the active layer of the photovoltaic device.

Conclusion and Outlook. We have achieved high performing OPVs exhibiting state of the art eciencies with the PTB7:PC₇₁BM system using a TAT multilayer electrode as replacement for ITO. Our TAT multilayers show a superior thermal stability than T/ITO, making them viable for application in other photovoltaic systems like DSSCs

Conclusion and Outlook. We have achieved high performing OPVs exhibiting state of the art eciencies with the PTB7:PC₇₁BM system using a TAT multilayer electrode as replacement for ITO. Our TAT multilayers show a superior thermal stability than T/ITO, making them viable for application in other photovoltaic systems like DSSCs